Working Memory Games Explained

Working Memory Test reveals grid positions one at a time and expects the same order. Memory Sequence uses a small set of colors that can repeat. In both, selecting the right items in the wrong order is an error because serial position is part of the rule.

A level may increase after success and fall or reset after an error. That staircase is a page policy, not a natural unit shared by every sequence task. Compare level only with the same cue timing, board, progression rule, and response method.

Visual Memory briefly marks several locations together, hides them, and accepts the target locations in any response order. The outcome concerns a set of positions rather than a presented sequence. Added non-target positions and omitted target positions should remain separate because they reflect different response mistakes.

Board dimensions, cell size, exposure duration, and whether the user changes zoom affect the task. A level from this page should not be compared numerically with a digit length or color sequence length.

At each step of an n-back task, the current item is compared with the item exactly n positions earlier. The site's Dual N-Back uses separate visual-position and letter channels. A position match and a letter match can occur independently, so the user maintains two response rules while the stream continues.

Hits are correctly reported matches, misses are matches not reported, and false alarms are reported matches that did not exist. Accuracy without those counts can conceal a response strategy. The n-level changes comparison distance and cannot be read as a number of ordinary items remembered.

A span page may report the last completed length. A sequence page may report a level and strict full-sequence success. Visual Memory can score target locations against omissions and additions. N-back accuracy depends on match opportunities across channels. Even when two pages show a percentage, the counted events differ.

Keep raw counts where available and read the scoring explanation before interpreting progress. Never average levels from different tasks. If a combined number has no defined rule, denominator, and validation, it is decoration rather than a meaningful measurement.

Repeated sessions teach cue locations, response mappings, timing, grouping, and recovery after mistakes. Improvement on the practiced task can be genuine while remaining specific to that task. Demonstrating transfer requires untrained outcomes and suitable control conditions, not only a rising local history.

Published reviews of working-memory and n-back training have found that trained-task or near-task gains do not justify confident claims about intelligence or broad real-world abilities. This site therefore reports page behavior and does not promise that Dual N-Back raises IQ.

Choose Number Memory for typed digit strings, Memory Sequence for repeating color order, Working Memory for ordered grid locations, Visual Memory for an unordered location set, and Dual N-Back for continuous comparison across visual and letter channels. Pick one mechanism rather than stacking every task.

For personal comparison, preserve cue timing, difficulty, device, viewport, input, and progression settings. Stop when the planned block ends. A longer session is not automatically more informative, particularly when fatigue or loss of interest changes the response strategy.